AI Engineer Coach

AI Engineer Coach analyzes how you actually use AI coding assistants by reading local session logs — from any harness, in one dashboard — rather than requiring you to adopt a specific tool’s built-in analytics. It tracks practice scores, weekly trends, and daily activity, but its more distinctive feature is anti-pattern detection: 45 rules spanning prompt quality, session hygiene, code review habits, tool mastery, and context management flag habits worth improving.

Beyond tracking, it measures AI-generated code output by language, workspace, model, and harness, and specifically looks for repeated prompts that could be turned into reusable skills — surfacing patterns you might not notice yourself across dozens of sessions. Context health scoring audits instruction files and maps workspace context to assess how “agentic-ready” a given codebase actually is for AI-assisted work.

MIT licensed and built by Microsoft, AI Engineer Coach processes everything locally — no data leaves your machine — and targets VS Code 1.115+ as a native extension rather than a separate application.

What You Get

• Progress tracking with practice scores, weekly trends, and daily activity charts
• Anti-pattern detection across 45 rules covering prompt quality, session hygiene, code review, tool mastery, and context management
• AI-generated code output measurement broken down by language, workspace, model, and harness
• Context health scoring, including instruction-file audits and workspace context mapping

Common Use Cases

• Identifying recurring bad habits in how you prompt or review AI-generated code across many sessions
• Measuring how much AI-generated code a project actually contains, broken down by model or tool used
• Discovering repeated prompts that would be better turned into reusable, saved skills
• Auditing a workspace’s instruction files and context setup to assess how AI-agent-ready it actually is

Under The Hood

Architecture AI Engineer Coach reads session logs directly from wherever local AI coding tools store them, rather than requiring a specific tool’s API or plugin hook — the “any harness, one dashboard” framing implies a log-parsing layer that normalizes different tools’ session formats into one common analysis model. The 45-rule anti-pattern engine and context-health scoring both operate on that normalized session/workspace data to produce actionable, categorized feedback rather than raw usage statistics alone.

Tech Stack TypeScript, packaged as a VS Code extension (targeting VS Code 1.115+), processing session logs and workspace files entirely locally with no network calls sending data externally.

Code Quality Very active, consistently maintained commit history and Microsoft’s backing lend credibility to the project’s data-handling claims; as a relatively young project, longer-term maturity signals like extensive third-party review are still developing.

What Makes It Unique Most AI coding analytics tools track usage volume or cost; AI Engineer Coach specifically targets behavioral improvement — detecting anti-patterns in how you prompt, review, and manage context — and closes the loop by surfacing repeated prompts as candidate reusable skills, treating AI-assisted coding as a skill to be coached rather than just metered.

Self-Hosting

Licensing Model MIT licensed — fully open source with no license key.

Self-Hosting Restrictions Not applicable; it’s a local VS Code extension with no hosted service, and explicitly processes all data on-device.

License Key Required No.